Classifications

• • G—PHYSICS
  • G02—OPTICS
  • G02C—SPECTACLES; SUNGLASSES OR GOGGLES INSOFAR AS THEY HAVE THE SAME FEATURES AS SPECTACLES; CONTACT LENSES
  • G02C13/00—Assembling; Repairing; Cleaning
  • G02C13/003—Measuring during assembly or fitting of spectacles
• • G—PHYSICS
  • G02—OPTICS
  • G02C—SPECTACLES; SUNGLASSES OR GOGGLES INSOFAR AS THEY HAVE THE SAME FEATURES AS SPECTACLES; CONTACT LENSES
  • G02C13/00—Assembling; Repairing; Cleaning
  • G02C13/003—Measuring during assembly or fitting of spectacles
  • G02C13/005—Measuring geometric parameters required to locate ophtalmic lenses in spectacles frames
• • G—PHYSICS
  • G02—OPTICS
  • G02B—OPTICAL ELEMENTS, SYSTEMS OR APPARATUS
  • G02B27/00—Optical systems or apparatus not provided for by any of the groups G02B1/00 - G02B26/00, G02B30/00
  • G02B27/0093—Optical systems or apparatus not provided for by any of the groups G02B1/00 - G02B26/00, G02B30/00 with means for monitoring data relating to the user, e.g. head-tracking, eye-tracking
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
  • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
  • G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
  • G06F3/012—Head tracking input arrangements
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
  • G06N3/047—Probabilistic or stochastic networks
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
  • G06Q30/00—Commerce
  • G06Q30/02—Marketing; Price estimation or determination; Fundraising
  • G06Q30/0201—Market modelling; Market analysis; Collecting market data
  • G06Q30/0206—Price or cost determination based on market factors
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
  • G06T17/10—Constructive solid geometry [CSG] using solid primitives, e.g. cylinders, cubes
• • H—ELECTRICITY
  • H04—ELECTRIC COMMUNICATION TECHNIQUE
  • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
  • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
  • H04N13/20—Image signal generators
  • H04N13/282—Image signal generators for generating image signals corresponding to three or more geometrical viewpoints, e.g. multi-view systems

Definitions

• the invention relates to a computer-implemented method for determining a representation of a spectacle-edge or a representation of the edges of the glasses of spectacles according to the preamble of claim 1.
• Methods are known which find the corresponding contours in image recordings for a given tracer record.
• Image-based methods such as pattern search or template matching are used. Reference is made by way of example to the disclosure of DE 10 201 1 1 15 239 B4.
• Tracer data is usually only done in the evening when the customer has already left the store, so he does not have to wait. In the prior art, therefore, only tracer data would have to be generated before the centering can be performed, which would change the workflow of the optician and cost the customer time. Additionally, it is also awkward when every optician is too each version must have the tracer record in stock. These data are usually only two-dimensional, but are usually required three-dimensional. It is therefore an object of the invention to develop a method of the type mentioned in such a way that it is more versatile applicable.
• the invention is based on the idea, the frame edge of a pair of glasses on the head of a subject who has been recorded in at least two mutually calibrated recordings, based on a parametric
• calibrated images are provided.
• Their calibration includes the extrinsic properties of the cameras taking the pictures or the camera taking pictures in succession, such as the relative orientation of their optical axes and the relative arrangement of each other in space, as well as their intrinsic properties, ie the characteristics of the cameras themselves, which define how Point in the space, which is located in the internal coordinate system of the respective camera, on which the coordinates of the pixels of the recorded image are mapped.
• intrinsic properties ie the characteristics of the cameras themselves, which define how Point in the space, which is located in the internal coordinate system of the respective camera, on which the coordinates of the pixels of the recorded image are mapped.
• a three-dimensional model of the spectacle frame or spectacles is provided which is based on geometrical parameters that define the shape of the spectacle frame or spectacles. These geometric parameters are optimized to match the model to the detected edges, and changing the parameters results in a change in the shape of the eyeglass frame or goggles. In addition, it is possible to optimize further geometric parameters,
• the representation of the spectacle frame edge or the representation of the edges of the glasses of the spectacles is usually used to determine
• Centering parameters used. Centering parameters are used to correctly arrange or center lenses in a spectacle frame, so that the
• Eyeglass lenses are arranged in the correct position relative to the eyes of the person wearing the glasses. These are partly anatomical
• Parameters of the person concerned such as the pupil distance, in part to purely barrel-specific parameters such as the lens width or the mounting disc height and partly to combinations of anatomical and specific parameters, such as the
• the images consist of pixels and that each pixel be assigned a mapping value representing the probability that the pixel will contain a portion of the edges of the spectacle frame or lenses.
• the detection of the frame edges or the glass edges is simplified.
• a probability map mapping these probabilities may be e.g. using edge detectors (gradient filters, Canny Edge Detector, Laplace
• edges of the spectacle frame or lenses can also be detected in the individual images, e.g. using Hough transformation, template matching or
• the spectacle frame or the spectacles can already be well defined in a first approximation.
• Another simplification of the model is that it defines each edge of the frame or each glass approximately as lying in one plane.
• the glasses may also be heated by means of higher order surfaces, e.g.
• concrete representations of the spectacle frame or the lenses are generated from the model by selecting different model parameter sets and projected onto the recorded images.
• these representations depict striking areas of the spectacle frame or the glasses, such as the glass edges or nasal and / or temporal edges.
• the projection is for the purpose of optimizing the parameters of the model so that the model of the captured images and the data generated therefrom, in particular the o.a. the probability of imaging values or the previously detected frame or glass edges fits.
• the geometric parameters of the model are preferably optimized by using a cost function stochastic or deterministic.
• Cost functions are also called
• the cost function evaluates the correspondence of the representations projected on the recorded images. The corresponding deviations then flow into the cost function. The value determined by the cost function, referred to below as the cost value, can then be optimized.
• Cost functions are usually used deterministic optimization methods such. Gradient descent methods, simplex methods (e.g.
• Nelder-Mead method differential evolution methods, primal-dual approaches, graphene-theoretical methods or discrete graph-cut methods.
• a deterministic optimization method is free of random influences, unlike a stochastic optimization method, and calculates the same solution with the same starting value each time.
• parameters can be determined from the image data and incorporated into the model. For example, by means of geometric positioning, in particular a
• Triangulation method can be used, the position of the eyes are determined in space and it can thereby influence the position of the eye defining parameters in the model.
• the model can be simplified by restricting and / or limiting the number of geometric parameters by specifying assumptions about the geometry of the spectacle frame or lenses and / or specifying assumptions or statistics regarding face or frame features in the range of values. For example, one can start from a symmetrical with respect to a median plane geometry of the spectacle frame or glasses. It is also possible to use a previously determined center line between the glasses of the spectacles in the images in order to ensure the symmetry of the model, for example by using the
• Approximated glasses must project onto this centerline.
• there may be statistics about centering parameters from which statistics about the geometric parameters of the present model may be derived e.g. Corneal vertex distance statistics can help narrow the positioning of the socket to a small area in the room at a certain distance from the eyes.
• Preload angles help narrow the range of values of individual parameters of the model so that optimizing the cost function is simplified by either optimizing only over parameters within a certain typical value range or by directly incorporating the probability distribution over the parameter value range into the optimization (eg Markov's Chain-Monte-Carlo method).
• the model may include a probability distribution that
• Probability value for a parameter below a predetermined threshold it may be provided to discard this parameter determined in the method to the dimension of the model, i. its complexity, too
• the computer-implemented method according to the invention is carried out with a device as described in principle in claim 15 and in detail in the following description of the figures.
• the representation of the edge of the spectacle lens or the representation of the spectacle frame edge determined in accordance with the invention becomes advantageous for the determination of Centering parameters used, the thus determined
• Spectacle frame are used.
• the at least one spectacle lens is centered with the specific centering parameters in the spectacle frame, or the at least one spectacle lens is ground on the basis of the determined centering parameters for an arrangement in the spectacle frame. In this way, lenses and glasses can be made.
• Figure 1 a, b is a device for determining Zentrierparametern in
• Figure 2a, b views of a wearing a spectacles head with projected on this approaching frame edges from the front and from the side and
• FIG. 3 shows a flow chart for illustrating a method according to a preferred exemplary embodiment.
• the device 10 shown in the drawing is used to determine Zentrierparametern for spectacle fitting. It has a column 12 which carries a rigid camera support 14 which is height adjustable and in turn carries a number of cameras 16a, 16b.
• the camera carrier 14 is in plan view
• Camera Carrier 14 an interior space 22 in which the head of a subject is when taking pictures by the cameras 16a, 16b.
• the inner surface 20 is concavely bent in a direction extending between the free ends 18 and has, for example, the shape of a portion of a Cylinder surface on which the cylinder is a circular or oval
• a not shown lifting device is arranged in the column 12, with the
• Camera carrier 14 can be moved by motor driven up and down.
• All cameras 16a, 16b are in one between the free ends 18th
• Embodiment is the camera assembly 26 as a camera series 26th
• the camera row 26 includes one in the middle of the
• Camera carrier 14 arranged front camera 16a, whose optical axis is directed frontally on the face of the subject, and eight pairs symmetrically with respect to a running through the optical axis of the front camera 16a perpendicular symmetry plane side cameras 16b of which four from the left and from the right to the face of the Subjects are addressed.
• the cameras 16a, 16b are also calibrated so that they can simultaneously take calibrated images of the subject. Calibration encompasses the extrinsic properties such as the relative orientation of their optical axes and the relative position of each other in space, as well as their intrinsic properties, ie the properties of the cameras themselves, which define how a point in the optical axis
• the camera support 14 encloses the interior 22 only forward, towards the column 12, and to the sides, so left and right of the head of the subject. Upwards, downwards and toward a rear side 30, it is open, the free ends 18 being at a distance of at least 25 cm from one another, so that the test person can easily approach from the rear side. Im shown
• a lighting device is provided with an upper light bar 32 extending above the camera row 26 and a lower light bar 34 extending below the camera row 26, which each have a multiplicity of LEDs as lighting means.
• the upper light bar 32 and the lower light bar 34 each extend continuously or intermittently over a length which is at least as long as the length of the circumferentially between the free ends 18 measured length of the camera row 26. This corresponds to a circumferential angle of at least 160 degrees , Near the free ends 18, the upper light bar 32 and the lower light bar 34 are each connected to each other by means of a vertically extending further light bar 36.
• the camera row 26 is thus completely framed by at least one row of LEDs.
• the device 10 also has a control or regulating device, not shown in detail in the drawing, with which the light intensity emitted by the LEDs depends on the light emitted by the
• Cameras 16a, 16b detected light intensity can be controlled or regulated.
• the LEDs of the light strips 32, 34, 36 are to sectors
• the radiated light intensities can be controlled or regulated separately.
• the light intensities emitted by the individual LEDs can also be controlled or regulated separately from one another by means of the control or regulating device.
• the two side cameras 16b closest to the front camera 16a are arranged to adjust the distance of the subject's head from the center 38 of the patient
• Camera carrier 14 to measure.
• Display unit is displayed to the subject whether he is correct or not.
• the display unit has a plurality of differently colored light sources, which are arranged in a row.
• the middle light source glows green when the subject is correct. Starting from the middle light source, there are one yellow, one orange and one red in each direction in this order
• Light source which indicates the color when the subject a little, clear or too far away from the center 38 of the camera support 14 or a little, clear or too close to the middle 38 stands.
• the camera carrier 14 In order to ensure that the viewing direction of the subject is infinite when determining the centering parameters, one is arranged on the camera carrier 14
• Fixation device 42 is provided which generates a fixation pattern for the subject in the form of a speckle pattern.
• the fixation pattern is arranged slightly higher than the front camera 16a, so that the subject on this
• the device 10 is also particularly suitable for producing an avatar of the subject's head, which can be used to determine the centering parameters.
• images calibrated by the cameras 16a, 16b of the subject's head are taken without glasses or spectacle frame.
• Position determination such as triangulation creates a depth profile of the head that approximates it very well.
• the head is represented by a multitude of points, which can be connected to each other by means of a net pattern or can be stored as a point cloud.
• the centering parameters of the thus determined avatar can be used to determine Zentrierparameter that due to the geometric properties of the glasses or
• Spectacle frame worn by the subject can not or only approximately can be determined.
• a wide frame strap can obscure the eye in a side view to such an extent that the corneal vertex distance can not or only very inaccurately be determined.
• colored or highly reflective glasses can not or only very poorly recognize the eyes.
• the depth profile of the avatar is projected onto the images taken by the cameras 16a, 16b of the subjects wearing the spectacles or spectacle frame, and the centering parameters, which can only be determined insufficiently due to the view restricted by the spectacles or spectacle frame, become determined by the image data of the avatar. It can be a Adapt the avatar to the pictures of the glasses or frame
• the apparatus 10 may be as follows to perform a computer-implemented method for
• Side cameras 16b images of a subject from the front ( Figure 2a) and from the side ( Figure 2b) taken, the subject wearing a pair of glasses or a
• Frame or glass edge belong.
• the parameters of the model are evaluated by means of a cost function and in a subsequent
• the model may initially start from a geometry of the mount or the eyeglasses, as described in FIGS. 2a, 2b by means of temporal and nasal edges 50, 52. Approximately, by means of each temporal edge 50 and the associated nasal edge 52, as illustrated in FIG. 2 a, a plane of glass can be laid which in the model approximates the spectacle lens.
• MCMC Markov chain Monte Carlo
• two planes are selected which are symmetrical to a median plane of the frame or spectacles.
• additional information such as the glass boxes in the front image, probability maps for the edges in the
• Edge detector and 3D position of the eyes given (step 60). From a set of training data (e.g., from a variety of known
• Eyewear orders are known Zentrierparameter for a larger amount of glasses. From these training data can be statistics about the
• Centering parameters are calculated.
• the statistics of the centering parameters can be found taking into account the additional geometric specifications
• the edges of the three-dimensional glass boxes 54 can now be projected into the page images (step 64).
• the projected edges can then be compared with the edges actually detected in the page image, e.g. using machine learning (Random Forest, Deep Learning, etc.) and distance metrics (e.g., Chamfer Distance). The better the projected edges match the detected ones, the higher the distance metrics (e.g., Chamfer Distance).
• This result can be rejected (step 68), for example if the probability of the sampled parameters or the sampled model and / or the score are too low.
• the MCMC algorithm works by generating a large number of level samples that simulate the model's probability density function (based on parameter likelihood and score). Especially many samples occur where parameter probability and score are high.
• the modes of this distribution eg with the mean shift algorithm (step 70).
• the centering parameters and the edges matching the selected model can then be read directly from the result parameter set in the page images.
• other optimization methods can also be used.
• the probability distribution over the parameters for finding a suitable starting value eg average or median of the distribution
• suitable limit values for the parameters to be optimized can be used.

Landscapes

• Physics & Mathematics (AREA)
• Engineering & Computer Science (AREA)
• General Physics & Mathematics (AREA)
• Theoretical Computer Science (AREA)
• Business, Economics & Management (AREA)
• Strategic Management (AREA)
• Finance (AREA)
• Geometry (AREA)
• Accounting & Taxation (AREA)
• Development Economics (AREA)
• Optics & Photonics (AREA)
• General Engineering & Computer Science (AREA)
• Health & Medical Sciences (AREA)
• Entrepreneurship & Innovation (AREA)
• Ophthalmology & Optometry (AREA)
• Software Systems (AREA)
• Data Mining & Analysis (AREA)
• Economics (AREA)
• Human Computer Interaction (AREA)
• Game Theory and Decision Science (AREA)
• Computer Graphics (AREA)
• Marketing (AREA)
• General Business, Economics & Management (AREA)
• Multimedia (AREA)
• Signal Processing (AREA)
• Life Sciences & Earth Sciences (AREA)
• Probability & Statistics with Applications (AREA)
• Artificial Intelligence (AREA)
• Biomedical Technology (AREA)
• Biophysics (AREA)
• Computational Linguistics (AREA)
• Evolutionary Computation (AREA)
• General Health & Medical Sciences (AREA)
• Molecular Biology (AREA)
• Computing Systems (AREA)
• Mathematical Physics (AREA)
• Eyeglasses (AREA)

Priority Applications (4)

Applications Claiming Priority (2)

Related Child Applications (1)

Publications (1)

Family

ID=57909540

Family Applications (1)

Country Status (5)

Cited By (1)

Families Citing this family (3)

Citations (3)

Family Cites Families (19)

• 2017
  • 2017-01-27 EP EP17153557.8A patent/EP3355101B1/de active Active
• 2018
  • 2018-01-24 EP EP18703242.0A patent/EP3574370B1/de active Active
  • 2018-01-24 ES ES18703242T patent/ES2813399T3/es active Active
  • 2018-01-24 WO PCT/EP2018/051729 patent/WO2018138149A1/de active Search and Examination
  • 2018-01-24 CN CN201880022311.4A patent/CN110446967B/zh active Active
• 2019
  • 2019-07-25 US US16/522,101 patent/US10571721B2/en active Active

Patent Citations (3)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events